Title :
Local Subspace Classifier with Transformation Invariance for Appearance-Based Character Recognition in Natural Images
Author :
Higa, Keisuke ; Hotta, Seiji
Author_Institution :
Div. of Adv. Inf. Technol. & Comput. Sci., Tokyo Univ. of Agric. & Technol., Koganei, Japan
Abstract :
This paper presents an appearance-based scheme for recognition of characters in natural images. In our method, we combine a local subspace classifier (LSC) and transformation-invariance with tangent vectors. In addition, we use negative images of original ones as new training samples for achieving high accuracy. Experimental results on Chars74K and ICDAR03-CH datasets show that the performance of our method is comparable to those of the feature-based state of the art.
Keywords :
character recognition; image classification; learning (artificial intelligence); Chars74K dataset; ICDAR03-CH dataset; LSC; appearance-based character recognition; local subspace classifier; natural image recognition; tangent vectors; transformation invariance; Accuracy; Character recognition; Image recognition; Manifolds; Training; Transforms; Vectors; appearance-based; character recognition; transform-invariance;
Conference_Titel :
Document Analysis and Recognition (ICDAR), 2013 12th International Conference on
Conference_Location :
Washington, DC
DOI :
10.1109/ICDAR.2013.112
